Understanding longitudinal growth modeling: An applied illustrative example examining resilience and changes in mental health among children of alcoholics

Despite epidemiologists' familiarity with traditional analyses of change, e.g., hazard models, less proficiency with longitudinal growth modeling (LGM) exists. Rather than examining event occurrence, LGMs examine change across time. They can examine: 1) whether variation in an outcome exists across time; 2) the form of this variation, e.g., linear, nonlinear, etc.; 3) whether groups vary in their averages across time, e.g., treatment vs. control; and 4) whether similarity exist in the rate of change across groups. Thus, LGMs are particularly appropriate for public health and policy research. For example, one might examine the extent to which a policy reduced health disparities across time and whether the effects and rate of change differed across minority groups. This presentation seeks to familiarize attendees with the LGM from an applied perspective. Using a sample of children of alcoholics (COAs: n = 216), it illustrates analyses examining the association of psychological resilience and mental health. Research suggests resilience is associated with decreased internalizing symptomatology. To more fully explore this relationship, LGM explored changes in symptomatology across three assessments. Analyses demonstrate that: 1) internalizing increases linearly across time for resilient and non-resilient COAs; 2) resilient COAs demonstrate lower levels of internalizing than non-resilient peers; and 3) resilient COAs demonstrate a significantly lower rate of change. Although COAs' internalizing increases across time, it does so more slowly than non-resilient COAs. Results describe the basic statistical model, focus on interpretation, and extend explanation to public health policy research examples.